Julian (aqora)

Julian (aqora)

Software Engineer

Paris, France

Joined January 22, 2024

Karma 8

Linkedin profile of julianhttps://julian.dev/
Overview
Last submissions
Last comments
Last topics
Organizations
Julian (aqora)

3

Posted by julian •

Hi @peterys! You can set the submitting user by setting the entity key in the pyproject.toml
# ...

[tool.aqora]
type = "submission"
competition = "<competition_name>"
entity = "<organization_username>"

# ...
Julian (aqora)

2

Posted by julian •

Hi @quantotto ! Thanks for question. I don't know how matplotlib works in a headless environment like we run the evaluator in, but I can check for you. I'll also check to see if there's a way for aqora to quiet matplotlib while running aqora test for convenience, but for right now, the easiest thing to do I guess is just to comment out plt.show() or similar before running aqora test
Julian (aqora)

1

Posted by julian •

Hi @peterys and @quantotto! I've updated the aqora cli! Try running pip install --upgrade aqora-cli or pipx upgrade aqora-cli or python -m pip install --upgrade aqora-cli (depending on how you installed the aqora-cli) and see if that helps, and let me know if you have any issues. Thanks for the patience
Julian (aqora)

2

Posted by julian •

Hi Matthew! You may want to check that you are running the kernel inside the virtual environment in VS code. In VS code when you have the notebook open you can select the kernel in the top right. It should be labeled .venv but if its not you can click on it and Select Another Kernel... > Python Environments > .venv (it should be starred). Let me know if that helps
Julian (aqora)

1

Posted by julian •

That helps a lot! I'll take a look and get back to you. Thanks!!
Julian (aqora)

2

Posted by julian •

Hi Peter! I'm really sorry for the slow response. I haven't been able to reproduce it yet unfortunately. How are you running the notebook locally? In VS Code? Could you give me your operating system, python version and the CLI version please! Thank you!!
Julian (aqora)

2

Posted by julian •

Hi Peter! I unfortunately don't know too much about D-Wave since I personally have never worked with it in the past. Have you reached out to Laia? I know she uses it quite a bit. I also know that Ingenii has a pretty active community you can reach out to as well https://www.ingenii.io/community
Julian (aqora)

1

Posted by julian •

Hi Peter! Me again! So I reached out to the competition organization and this is what they said:
His solution is totally right, that is a typo! I realized that the d = 0.0976 value would be considering only the first covariate w1, not the three of them.
Considering the three of them we have:
The solution arrays for this toy problem is the following:
  • group1 = [1, 0, 0, 1, 0, 1]
  • group2 = [0, 1, 1, 0, 1, 0]
The discrepancy value for this solution is d = 1.485
So I've gone ahead and updated the docs with this information. Sorry for the confusion! and I hope that helps answer your question.
Julian (aqora)

1

Posted by julian •

Hi Peter! Yeah, you're totally right. Let me check with host, to see how the inconsistency came about
Julian (aqora)

2

Posted by julian •

Hi Peter! So we take the data from here https://aqora.io/competitions/ingenii-clinical-trial/code/data/blob/pbc.csv and randomly select 100 people from the trial, but we then normalize the covariates using the following
(w - np.mean(w, axis=0)) / np.std(w, axis=0)
I hope that answers your question